Masters Theses

Abstract

Accurate and automated recognition of coded targets is crucial for high-precision photogrammetry-based measurements in triaxial testing. Traditional recognition methods, including those based on deep learning and table decoding, often struggle with issues such as perspective distortion, rotation, and variable lighting, leading to unreliable results. While deep learning approaches offer some improvements, they remain computationally intensive and sensitive to environmental factors.

This thesis introduces an innovative system that replaces deep learning algorithms with a blob analysis-based method for efficient and robust recognition of coded targets. The system also employs newly designed solid points of varying sizes and patterns, eliminating the need for complex algorithms like RANSAC. Through renumbering and interpolation techniques, the method enhances the coverage and spatial resolution of coded targets, enabling more accurate and detailed 3D reconstructions. Experimental validation confirms its superior performance in speed and accuracy, offering a more reliable and scalable solution for tracking soil deformation during triaxial tests.

Advisor(s)

Zhang, Xiong
Yan, Guirong Grace

Committee Member(s)

Wang, Jianmin

Department(s)

Civil, Architectural and Environmental Engineering

Degree Name

M.S. in Civil Engineering

Publisher

Missouri University of Science and Technology

Publication Date

Summer 2025

Pagination

vii, 67 pages

Note about bibliography

Includes_bibliographical_references_(pages 60-63)

Rights

© 2025 Qingqing Fu , All Rights Reserved

Document Type

Thesis - Open Access

File Type

text

Language

English

Thesis Number

T 12537

Download

Share

 
COinS